Computer Science PhD School Visits, Revisited

Here you are, PhD admissions letters in hand, trying to decide how best to spend the next n years of your life. Congratulations! The part of the process that is out of your control is over. The hard work is about to begin.

I previously wrote a couple of other blog posts (School Visits; Somes notes on picking grad schools/advisors) about the process of choosing schools and advisors. Now that I have the wisdom and authority of someone about to start a faculty position, here is some revised and reorganized advice.

This is a pretty high-level post. For more concrete advice, you may find it helpful to read my previous post or Neha Narula’s blog post on specific questions to ask.

What I Would Consider Now

If I had to make the decision again about where to go for a PhD, here’s how I would go about considering the decision, in order of most important to least.

Advisor fit. When I attended graduate student orientation, I was told that my relationship with my advisor was going to be my most important relationship for the next however many years. This was true. You’re going to graduate school to grow into the researcher you’re going to be and your advisor will be the primary person overseeing your growth. You are going to go through some tough times with your advisor, so you need to start out with enough goodwill and respect to make it through. As with a business partnership or long-term romantic relationship, you will want to consider the following things:

• Compatible core values. Whether you like it or not, you’re going to pick up a lot from your advisor about how they are as a researcher, from the way they choose problems to the way they execute on problem to the way they walk and talk. It’s going to be a much easier growth process for you if you match on the same core values as your advisor, for instance how much you value formalism and how much you value impact. It is important to have compatible working styles. You’re going to do a lot of work with your advisor. It will really help if it’s not a battle every time you try to schedule a meeting or work towards a deadline. How hands-on is the advisor in managing the research process? How involved are they in the technical research? How involved are they in the paper-writing process? How much do they do things last minute, versus doing things early? How detail-oriented are they? How detail-oriented would they like you to be?
• How much they care about you. When you are starting out, especially if you want a lot of attention, it is good to find an advisor who is excited about you. Excitement often doesn’t last throughout the course of a PhD, so for your long-term happiness it’s helpful to get an advisor who cares about your career development, even if your interests veered out of their comfort zone. Part of this has to do with how well your interests and skills match what the advisor is looking for. Part of this has to do with how open the advisor is to the ideas of others and part of this has to do with how much they care about students in general. You can get a sense of this by talking to them and also to people they work with.
• "You marry the family." Especially if your advisor is more senior, you will spend more time with your groupmates and officemates from the department than with your advisor yourself. It is important to take this into account as part of the whole package.

Research area fit. You’re going to spend the most time talking to the people in your research area. They are the ones who will give you the most feedback on your choice in research problems and the way you go about solving them. The people in your research area will be who prepares you to communicate about and defend your research to the rest of the world. They are your team. You want to make sure you have one that will support you intellectually and otherwise in the ways you want. Also, it may turn out that you are not the best fit with your advisor. It is likely you’ll want to stay within the same area, so it’s good if you like the other professors in your area enough to be able to imagine working with some of them.

Department fit. Concretely, the requirements your department places on your PhD will affect your life. (How many courses do you have to take? How intense is the qual process? Are people in the department happy?) More abstractly, the department is your extended family. Presumably, you’re going to be friends with the other students in your department. Their research ideas and outlooks on life will influence your personal and professional development for the next few years. Also, you’re young: you might not stay forever in the research area that you discovered after taking a class with a charismatic professor your sophomore year. In case you decide your research area isn’t for you, it would be good to have the safety net of being in a department where you have the flexibility of moving around.

School fit. One of my favorite things about being in academia is about being embedded in a university. Being in a university means that you are surrounded by people passionate about intellectual and other pursuits. Each school has its own flavor and activities and it’s nice, if you are in a position to choose, to take this into consideration as well.

Lifestyle fit. Most people have things they like to do outside of work. By the time I was a senior undergraduate, I had spent so much of my time working I could not really imagine having hobbies again, but even I found space in my life for hobbies during my PhD. Each school will sell different lifestyle packages. Some schools are in cities. Some schools have good access to the outdoors. Some schools are in places where you could have an incredible quality of life (for a PhD student). Figure out what is important to you, whether the geographic location the school is in offers it, and whether the culture of the department and the resources (financial and otherwise) offered to you would allow you to achieve that lifestyle during your PhD years.

How I Would Consider Things

Here’s how I would go about making the decision given what I know now:

• Don’t go with your feelings. Most professors don’t get to become professors without learning how to sell--themselves, their research, and whatever else they need to sell to achieve their professorly goals. This selling usually involves a nontrivial amount of manipulating people’s emotions to make them feel positively. That’s right. Professors are masters at making you feel whatever they want to make you feel. From this we can derive that professors are masters at making you feel you should go to their school and work with them. Therefore, it’s dangerous to trust your feelings. I recommend allowing feelings to settle for as long as possible after talking to any professor before you make a decision.
• Talk to as many people as possible, especially people not invested in where you go. Many people are invested in you making a certain decision. Professors who are recruiting you want students--and like to win. Their students like validation that they chose the right advisor--and school. In general, people like to win. The more invested people are in your decision, the more they are going to be charming and make statements designed to make you feel more positively about a certain school or advisor.
• Go to a place that feels different from where you are now. Many people have an attraction to what feels “safe.” To overcome this it’s often important to place more weight on the new and different so that you have more opportunities to grow.

What I Actually Did

My undergraduate professor Radhika Napgal says to take all advice with a grain of salt, since it’s largely highlights and wishful thinking. To support her point, here is what I actually did. :)

When I was a college senior, I valued fit in this order: lifestyle, school, department, research area, advisor. If you want to make your decisions based on how I made mine, here is what to do: Talk to some people. Get really excited that all these senior professors you had idolized are actually talking to you. Get overwhelmed with how many people are telling you different things. Avoid talking to people after that. Out of lack of other strong heuristics, eliminate schools based on the geographic preferences of your significant other at the time. Realize this was not the best heuristic after you and said significant other break up a year later. Allow multiple local professors to take you out to meals. Disregard the fact that you may feel more positively about their school because you have received gifts of food and time from them. Go with which school “feels” like the best fit.

Based on the decision-making process above, I ended up choosing MIT, even though it wasn’t the best research fit. Not realizing I had some agency in deciding who would be my advisor, I ended up getting assigned Mike Ernst and Saman Amarasinghe. Mike Ernst left for UW the week I arrived. Saman and I got along very well and I loved talking to him and getting life advice from him, but it wasn’t a great research fit. My PhD advisor, Armando Solar-Lezama, showed up to MIT November of my first year. His PhD advisor, Ras Bodik, had tried to recruit me to Berkeley but I told him I felt MIT was a better fit personality-wise, despite my preference for Berkeley-style projects. Ras quoted this back to me and told me to work with Armando. I became Armando’s first student. Saman and I still hang out. I attended his group meetings for years after I quit the group. Last summer he invited me to Singapore with him and got the Singaporean government to fly me business class both ways.

Things turned out all right for me, but I got pretty lucky. From my story you may take comfort in the fact that it is possible to remain alive and in research even after engaging in questionable decision-making processes. To avoid making my mistakes, however, you may also want to read about the time I almost accidentally joined a cult.

Parting Words

Remember to have fun! If it’s a hard decision, then don’t worry so much about it. You probably can’t go wrong. Try not to accidentally join any cults.

Also. If you’re reading this and CMU has admitted you, I’d love for you to join us. I love how collegial CMU is and the breadth and depth of its School of Computer Science. (How many other places have so much CS that it constitutes a School?) A graduate stipend goes a long way in Pittsburgh, which is becoming more hip every day. (I’ve long thought it’s the next Portland, even before others started saying so.) Programming languages at CMU is particularly great because we have the largest group at a university that I know of, with an incredible range from theory to practice. It’s particularly active now that they hired Jan Hoffmann, Matt Fredrikson, and me. And let’s talk about why I’m the best advisor for you. ;)